To classify a tumor in digital mammogram images, I have tested several families of features like ho moments, harralik moments, co-occurrence matrix and walwet.
What are the most appropriate features?
Hayet Farida Djellali Merouani
Nabiha, we tested several techniques as you said, Haralick, co-occurence Matrix, wavelet, Randm Forest, Zipf 'Law, Inverse Zipf' Law, combination of agents and automata cat this moment we got a very good results with the combination of Zipf and Zipf Law better. Let's see what others say??
1 votes
0 thanks
Hayet Farida Djellali Merouani
well, it depends of your application and specially about your image type. How much it' s dense and the kind of the image texture;;;;
1 votes
1 thanks
- Badges
- Science topic
- Similar topics
- Biological Science
- Microbiology
- Virology
- Virus
More Nabiha Azizi's questions See All
Similar questions and discussions